Structural analysis and semantic understanding for offline mathematical expressions

As the Internet has been widely applied and the digital library developed, recognition and understanding for mathematical expressions that are included in scientific papers have become a significant research area. Here, we put forward a method to recognize and understand mathematical expressions based on a rule base and an overall design for this system. We enumerate some ambiguous problems in expression understanding based on the existing mathematical expressions found in some published books. Because understanding for the mathematical expression ambiguity has certain connections with people's sense and experience, we introduce mathematical rules, sense-based and experience-based dictionaries into the rule base. We take advantage of the parser based on mathematical rules in the rule base to understand expressions without ambiguity and utilize a combination strength function to understand expressions with ambiguity. As experimental results, layout and semantic trees for mathematical expressions are produced by making use of mathematical rules after we have decomposed two-dimensional expression structure notations into one-dimensional expressions. At the same time, some expressions with ambiguity are understood by computing and comparing combination strength among various symbols in the expressions.